Mirelo AI API Examples

This directory contains complete examples for using the Mirelo AI Audio Generation API in both Python and Node.js/TypeScript.

Prerequisites

• API Key from studio.mirelo.ai
• Python 3.8+ (with uv installed) or Node.js 18+
• FFmpeg installed (for video concatenation in example 3)

Setup

Configuration

Set your API key as an environment variable:

export MIRELO_API_KEY="your_api_key_here"

Examples

1. Text-to-SFX (No Video Required)

Generate sound effects from text descriptions only.

Python:

uv run python/1_text_to_sfx.py

Node.js:

npm run example1

What it does:

• Generates SFX from text description without any video input
• Uses the /video-to-sfx endpoint with text_prompt parameter
• Downloads generated audio files

---

2. Short Video SFX (≤10 seconds)

Generate sound effects for a short video with full parameter control.

Python:

uv run python/2_short_video_sfx.py

Node.js:

npm run example2

What it does:

• Uploads a video file (≤10 seconds)
• Generates SFX with custom parameters (duration, creativity, model version)
• Returns video with synchronized audio (set return_audio_only: false)
• Downloads the result video

---

3. Long Video SFX (>10 seconds, up to 10 minutes)

Generate sound effects for long videos using automatic scene detection.

Python:

uv run python/3_long_video_sfx.py

Node.js:

npm run example3

What it does:

• Uploads a video file (>10 seconds, up to 10 minutes)
• Uses automatic scene detection to split video into segments
• Generates SFX for each segment (requests 1 sample per scene)
• Downloads audio files for each segment
• Uses FFmpeg to concatenate the selected audio with the original video

---

API Endpoints Used

/create-customer-asset

Creates an upload URL for your video file.

/video-to-sfx

Generates sound effects for short videos or from text descriptions.

Key Parameters:

• text_prompt - Text description of desired audio (optional, can be used without video)
• duration - Duration in seconds (1-10)
• num_samples - Number of variations to generate (1-4)
• model_version - Model to use ("1.0" or "1.5")
• creativity_coef - Creativity level (1-10)
• return_audio_only - Return just audio (true) or video with audio (false)
• customer_asset_id - ID from uploaded video (optional)
• video_url - Public video URL (optional)

/long-video-to-sfx

Generates sound effects for long videos with automatic scene detection.

Key Parameters:

• num_samples - Number of audio variations per scene (1-8)
• customer_asset_id - ID from uploaded video
• video_url - Public video URL (alternative to customer_asset_id)

Returns: Array of video segments with start time, duration, and audio URLs for each segment.

---

Support

For questions or issues:

• Visit mirelo.ai
• View API docs at api.mirelo.ai/docs